Merge branch 'master' of https://git.d-popov.com/popov/neuron-tai

Combine shard label formatting with model/shard flag parsing and tracker registration retry.

Co-authored-by: Cursor <cursoragent@cursor.com>

.gitignore

@@ -19,4 +19,8 @@ dist/
 !.env.testnet
 .rocm-local/*
 billing.sqlite
-.pytest-tmp/*
+.pytest-tmp/*
+billing.sqlite
+meshnet_registry.sqlite3
+billing.sqlite
+meshnet_registry.sqlite3

packages/node/meshnet_node/model_backend.py

@@ -134,8 +134,9 @@ class TorchModelShard:
                     self.model.to(self.device)
         except Exception as exc:
             if _looks_like_oom(exc):
+                memory_kind = "VRAM" if self.device.type == "cuda" else "RAM"
                 raise InsufficientVRAMError(
-                    f"insufficient VRAM to load {model_id} layers {shard_start}:{shard_end} "
+                    f"insufficient {memory_kind} to load {model_id} layers {shard_start}:{shard_end} "
                     f"with {quantization} quantization; choose a smaller shard or lower quantization"
                 ) from exc
             raise
@@ -215,7 +216,7 @@ class TorchModelShard:
     def generate_text(
         self,
         messages: list[dict],
-        max_new_tokens: int = 256,
+        max_new_tokens: int = 5120,
         temperature: float = 1.0,
         top_p: float = 1.0,
     ) -> str:
@@ -245,7 +246,7 @@ class TorchModelShard:
     def generate_text_streaming(
         self,
         messages: list[dict],
-        max_new_tokens: int = 256,
+        max_new_tokens: int = 5000,
         temperature: float = 1.0,
         top_p: float = 1.0,
     ):
@@ -411,7 +412,7 @@ def _should_partial_materialize_shard(
         return False
     if total_layers_hint is None:
         return False
-    return not (shard_start == 0 and shard_end >= total_layers_hint - 1)
+    return True
 
 
 def _load_partial_model_from_snapshot(
@@ -476,7 +477,7 @@ def _load_partial_model_from_snapshot(
         )
 
     with init_empty_weights_fn():
-        model = auto_model_for_causal_lm.from_config(cfg, torch_dtype=dtype)
+        model = auto_model_for_causal_lm.from_config(_causal_lm_config(cfg), torch_dtype=dtype)
     tie_weights = getattr(model, "tie_weights", None)
     if callable(tie_weights):
         tie_weights()
@@ -498,7 +499,7 @@ def _load_partial_model_from_snapshot(
             for tensor_name in names:
                 set_tensor_fn(
                     model,
-                    tensor_name,
+                    _checkpoint_tensor_name_for_model(model, tensor_name),
                     device,
                     value=handle.get_tensor(tensor_name),
                     dtype=dtype,
@@ -569,38 +570,74 @@ def _native_torch_dtype(cfg: Any, torch: Any) -> Any:
     return torch.bfloat16
 
 
+def _causal_lm_config(cfg: Any) -> Any:
+    """Use the text decoder config for composite VLM/MoE presets."""
+    get_text_config = getattr(cfg, "get_text_config", None)
+    if callable(get_text_config):
+        try:
+            return get_text_config()
+        except Exception:
+            pass
+    text_config = getattr(cfg, "text_config", None)
+    if text_config is not None:
+        return text_config
+    return cfg
+
+
+def _checkpoint_tensor_name_for_model(model: Any, tensor_name: str) -> str:
+    """Map multimodal checkpoint keys onto text-only CausalLM modules when needed."""
+    inner = getattr(model, "model", None)
+    if inner is not None and hasattr(inner, "language_model"):
+        return tensor_name
+    if ".language_model." in tensor_name:
+        return tensor_name.replace(".language_model.", ".")
+    return tensor_name
+
+
+def _transformer_backbone(model: Any) -> Any:
+    if hasattr(model, "model"):
+        inner = model.model
+        language_model = getattr(inner, "language_model", None)
+        if language_model is not None:
+            return language_model
+        return inner
+    if hasattr(model, "transformer"):
+        return model.transformer
+    raise ModelBackendError(
+        "unsupported HuggingFace model architecture: no transformer backbone found"
+    )
+
+
 def _model_layers(model: Any) -> Any:
-    if hasattr(model, "model") and hasattr(model.model, "layers"):
+    backbone = _transformer_backbone(model)
-        return model.model.layers
+    for attr in ("layers", "h", "blocks"):
-    if hasattr(model, "transformer") and hasattr(model.transformer, "h"):
+        layers = getattr(backbone, attr, None)
-        return model.transformer.h
+        if layers is not None:
+            return layers
     raise ModelBackendError(
         "unsupported HuggingFace model architecture: no transformer layers found"
     )
 
 
 def _embed_tokens(model: Any) -> Any:
-    if hasattr(model, "model") and hasattr(model.model, "embed_tokens"):
+    backbone = _transformer_backbone(model)
-        return model.model.embed_tokens
+    for attr in ("embed_tokens", "wte"):
-    if hasattr(model, "transformer") and hasattr(model.transformer, "wte"):
+        embed = getattr(backbone, attr, None)
-        return model.transformer.wte
+        if embed is not None:
+            return embed
     raise ModelBackendError(
         "unsupported HuggingFace model architecture: no token embeddings found"
     )
 
 
 def _position_embeddings(model: Any) -> Any | None:
-    if hasattr(model, "transformer") and hasattr(model.transformer, "wpe"):
+    backbone = _transformer_backbone(model)
-        return model.transformer.wpe
+    return getattr(backbone, "wpe", None)
-    return None
 
 
 def _rotary_embedding_module(model: Any) -> Any | None:
-    if hasattr(model, "model") and hasattr(model.model, "rotary_emb"):
+    backbone = _transformer_backbone(model)
-        return model.model.rotary_emb
+    return getattr(backbone, "rotary_emb", None)
-    if hasattr(model, "transformer") and hasattr(model.transformer, "rotary_emb"):
-        return model.transformer.rotary_emb
-    return None
 
 
 def _active_modules_for_shard(model: Any, shard_start: int, shard_end: int) -> list[Any]:
@@ -627,10 +664,11 @@ def _active_modules_for_shard(model: Any, shard_start: int, shard_end: int) -> l
 
 
 def _final_norm(model: Any) -> Any | None:
-    if hasattr(model, "model") and hasattr(model.model, "norm"):
+    backbone = _transformer_backbone(model)
-        return model.model.norm
+    for attr in ("norm", "ln_f", "final_layer_norm"):
-    if hasattr(model, "transformer") and hasattr(model.transformer, "ln_f"):
+        norm = getattr(backbone, attr, None)
-        return model.transformer.ln_f
+        if norm is not None:
+            return norm
     return None
 
 
@@ -743,7 +781,12 @@ def _looks_like_oom(exc: BaseException) -> bool:
     current: BaseException | None = exc
     while current is not None:
         text = str(current).lower()
-        if "out of memory" in text or "cuda error: out of memory" in text:
+        if (
+            "out of memory" in text
+            or "cuda error: out of memory" in text
+            or "paging file is too small" in text
+            or "os error 1455" in text
+        ):
             return True
         current = current.__cause__ or current.__context__
     return False

packages/node/meshnet_node/relay_bridge.py

@@ -164,6 +164,9 @@ class RelayHttpBridge:
         path = str(payload.get("path") or "/")
         headers = payload.get("headers") if isinstance(payload.get("headers"), dict) else {}
 
+        req_suffix = f" request_id={request_id}" if request_id else ""
+        print(f"  [node] relay {method} {path}{req_suffix}", flush=True)
+
         # body_base64 carries binary data (e.g. bfloat16 activation tensors) safely.
         # Fallback to text "body" for backward-compat with non-binary requests.
         body_b64 = payload.get("body_base64")

packages/node/meshnet_node/startup.py

@@ -205,6 +205,21 @@ def _max_assignable_layers(
     return min(total_layers, int((budget_bytes * 0.8) // layer_bytes))
 
 
+def _format_shard_label(
+    shard_start: int,
+    shard_end: int,
+    total_layers: int | None = None,
+    *,
+    model_name: str | None = None,
+) -> str:
+    layer_count = shard_end - shard_start + 1
+    if isinstance(total_layers, int) and total_layers > 0:
+        return f"layers {shard_start}–{shard_end} ({layer_count} of {total_layers})"
+    if model_name:
+        return f"layers {shard_start}–{shard_end} ({model_name})"
+    return f"layers {shard_start}–{shard_end}"
+
+
 def _shard_budget_line(
     memory_mb: int,
     memory_source: str,
@@ -734,11 +749,7 @@ def run_startup(
         _node_start_time = time.monotonic()
         actual_port = node.start()
         total_layers = getattr(getattr(node, "backend", None), "total_layers", None)
-        if isinstance(total_layers, int) and total_layers > 0:
+        shard_label = _format_shard_label(shard_start, shard_end, total_layers)
-            layer_count = shard_end - shard_start + 1
-            shard_label = f"layers {shard_start}–{shard_end}; {layer_count} of {total_layers}"
-        else:
-            shard_label = f"layers {shard_start}–{shard_end}"
         public_host = advertise_host or (socket.getfqdn() if host == "0.0.0.0" else host)
         endpoint = f"http://{public_host}:{actual_port}"
         local_base_url = f"http://127.0.0.1:{actual_port}"
@@ -913,14 +924,17 @@ def run_startup(
         tracker_node_id = _register_with_tracker(
             tracker_url, auto_reg_payload, node, _node_start_time,
         )
-        shard_count = assigned_shard_end - assigned_shard_start + 1
+        shard_label = _format_shard_label(
+            assigned_shard_start,
+            assigned_shard_end,
+            assigned_num_layers,
+        )
         print(
             f"\n{'=' * 32}\n"
             f"meshnet-node ready (auto-joined)\n"
             f"  Wallet:       {address}\n"
             f"  Model ID:     {assigned_hf_repo}\n"
-            f"  Shard:        layers {assigned_shard_start}–{assigned_shard_end} "
+            f"  Shard:        {shard_label}\n"
-            f"({shard_count} of {assigned_num_layers})\n"
             f"  {_shard_budget_line(memory_budget_mb, memory_budget_source, assigned_num_layers, quantization)}\n"
             f"  Quantization: {quantization}\n"
             f"  Endpoint:     {endpoint}\n"
@@ -967,13 +981,35 @@ def run_startup(
     peers: list[dict] = assignment.get("peers", [])
     model_sources: list[dict] = [] if tracker_source_disabled else assignment.get("model_sources", [])
     assignment_bytes_per_layer = _assignment_bytes_per_layer(assignment, quantization)
+    model_layers_end = assignment.get("model_layers_end")
+    assigned_total_layers = (
+        int(model_layers_end) + 1
+        if model_layers_end is not None
+        else None
+    )
+    shard_label = _format_shard_label(
+        shard_start,
+        shard_end,
+        assigned_total_layers,
+        model_name=assigned_model,
+    )
     if user_pinned_shard:
-        print(
+        shard_label = f"{shard_label} (pinned)"
-            f"  Shard: layers {shard_start}-{shard_end} of {assigned_model} (pinned)",
+    if user_pinned_shard and assigned_total_layers and assignment_bytes_per_layer:
-            flush=True,
+        pinned_layers = shard_end - shard_start + 1
+        max_layers = _max_assignable_layers(
+            memory_budget_mb,
+            assigned_total_layers,
+            assignment_bytes_per_layer,
         )
-    else:
+        if pinned_layers > max_layers:
-        print(f"  Shard: layers {shard_start}-{shard_end} of {assigned_model}", flush=True)
+            raise ValueError(
+                f"Pinned shard layers {shard_start}–{shard_end} ({pinned_layers} layers) exceed "
+                f"the {memory_budget_mb / 1024:.1f} GB {memory_budget_source} budget "
+                f"(fits up to {max_layers}/{assigned_total_layers} layers at bfloat16). "
+                "Drop --shard-start/--shard-end to let the tracker auto-assign, or pin a smaller range."
+            )
+    print(f"  Shard: {shard_label}", flush=True)
 
     # 4. Download shard
     print("Downloading shard...", flush=True)
@@ -998,7 +1034,77 @@ def run_startup(
     )
     print(f"  Cached at: {shard_path}", flush=True)
 
-    # 5. Start HTTP server
+    # 5. Start HTTP server — real HF weights use TorchNodeServer; stub-model stays stub.
+    _node_start_time = time.monotonic()
+    if hf_repo and assigned_model != "stub-model":
+        print("Loading real PyTorch model shard...", flush=True)
+        node = TorchNodeServer(
+            host=host,
+            port=port,
+            model_id=hf_repo,
+            shard_start=shard_start,
+            shard_end=shard_end,
+            quantization=quantization,
+            tracker_url=tracker_url,
+            route_timeout=route_timeout,
+            cache_dir=shard_path,
+            debug=debug,
+            max_loaded_shards=max_loaded_shards,
+        )
+        actual_port = node.start()
+        total_layers = getattr(getattr(node, "backend", None), "total_layers", None) or assigned_total_layers
+        shard_label = _format_shard_label(shard_start, shard_end, total_layers, model_name=assigned_model)
+        if user_pinned_shard:
+            shard_label = f"{shard_label} (pinned)"
+        public_host = advertise_host or (socket.getfqdn() if host == "0.0.0.0" else host)
+        endpoint = f"http://{public_host}:{actual_port}"
+        local_base_url = f"http://127.0.0.1:{actual_port}"
+        relay_bridge, relay_fields = _start_relay_bridge_if_available(
+            tracker_url,
+            address,
+            local_base_url,
+            endpoint,
+            relay_url=relay_url,
+        )
+        _attach_relay_bridge(node, relay_bridge)
+        reg_payload = {
+            "endpoint": endpoint,
+            "model": assigned_model,
+            "hf_repo": hf_repo,
+            "num_layers": total_layers,
+            "shard_start": shard_start,
+            "shard_end": shard_end,
+            "downloaded_models": downloaded_models,
+            "hardware_profile": hw,
+            "wallet_address": address,
+            "quantization": quantization,
+            "score": 1.0,
+            "tracker_mode": (shard_start == 0),
+            "managed_assignment": not user_pinned_shard,
+            "model_metadata": model_metadata_for(hf_repo, total_layers, cache_dir=shard_path),
+            **registration_capabilities,
+            **relay_fields,
+        }
+        tracker_node_id = _register_with_tracker(
+            tracker_url, reg_payload, node, _node_start_time,
+        )
+        print(
+            f"\n{'=' * 32}\n"
+            f"meshnet-node ready\n"
+            f"  Wallet:       {address}\n"
+            f"  Model ID:     {hf_repo}\n"
+            f"  Shard:        {shard_label}\n"
+            f"  {_shard_budget_line(memory_budget_mb, memory_budget_source, total_layers, quantization, bytes_per_layer=assignment_bytes_per_layer)}\n"
+            f"  Quantization: {quantization}\n"
+            f"  Endpoint:     {endpoint}\n"
+            f"  Node ID:      {tracker_node_id or 'unregistered'}\n"
+            f"  Hardware:     {_hardware_label(device, gpu_name)}\n"
+            f"  Benchmark:    {bench_tps:,.0f} (throughput index)\n"
+            f"{'=' * 32}",
+            flush=True,
+        )
+        return node
+
     is_last = shard_end >= assignment.get("model_layers_end", shard_end)
     node = StubNodeServer(
         host=host,
@@ -1009,7 +1115,6 @@ def run_startup(
         model=assigned_model,
         shard_path=shard_path,
     )
-    _node_start_time = time.monotonic()
     actual_port = node.start()
     public_host = advertise_host or (socket.getfqdn() if host == "0.0.0.0" else host)
     endpoint = f"http://{public_host}:{actual_port}"
@@ -1055,12 +1160,18 @@ def run_startup(
     hw_str = device.upper()
     if gpu_name:
         hw_str += f" ({gpu_name}, {vram_mb / 1024:.1f} GB)"
+    shard_label = _format_shard_label(
+        shard_start,
+        shard_end,
+        assigned_total_layers,
+        model_name=assigned_model,
+    )
     print(
         f"\n{'=' * 32}\n"
         f"meshnet-node ready\n"
         f"  Wallet:   {address}\n"
-        f"  Shard:    layers {shard_start}-{shard_end} ({assigned_model})\n"
+        f"  Shard:    {shard_label}\n"
-        f"  {_shard_budget_line(memory_budget_mb, memory_budget_source, assignment.get('model_layers_end', shard_end) + 1, quantization, bytes_per_layer=assignment_bytes_per_layer)}\n"
+        f"  {_shard_budget_line(memory_budget_mb, memory_budget_source, assigned_total_layers, quantization, bytes_per_layer=assignment_bytes_per_layer)}\n"
         f"  Endpoint: {endpoint}\n"
         f"  Node ID:  {node_id}\n"
         f"  Hardware: {hw_str}\n"

packages/node/meshnet_node/torch_server.py

@@ -113,6 +113,10 @@ class _TorchHandler(http.server.BaseHTTPRequestHandler):
     def log_message(self, fmt, *args):  # noqa: suppress request logs in tests
         pass
 
+    def _request_log_suffix(self) -> str:
+        req_id = self.headers.get("X-Meshnet-Request-Id") or self.headers.get("X-Request-Id")
+        return f" request_id={req_id}" if req_id else ""
+
     def do_POST(self):
         server: _TorchHTTPServer = self.server  # type: ignore[assignment]
         if self.path == "/forward":
@@ -199,8 +203,18 @@ class _TorchHandler(http.server.BaseHTTPRequestHandler):
             return
 
         server.forward_chunk_count += 1
-        if int(self.headers.get("X-Meshnet-Hop-Index", "0")) > 0:
+        hop_index = int(self.headers.get("X-Meshnet-Hop-Index", "0"))
+        if hop_index > 0:
             server.received_activations = True
+        if chunk_index_value == 0:
+            shard_start = getattr(server.backend, "shard_start", "?")
+            shard_end = getattr(server.backend, "shard_end", "?")
+            print(
+                f"  [node] forward hop={hop_index} "
+                f"layers={shard_start}-{shard_end} "
+                f"session={session[:8]}{self._request_log_suffix()}",
+                flush=True,
+            )
 
         start_layer_header = self.headers.get("X-Meshnet-Start-Layer")
         start_layer = int(start_layer_header) if start_layer_header else None
@@ -307,24 +321,57 @@ class _TorchHandler(http.server.BaseHTTPRequestHandler):
         if backend is None or not backend.is_head:
             self._send_json(400, {"error": "model not loaded on this node"})
             return
-        max_tokens = int(body.get("max_tokens") or body.get("max_new_tokens") or 256)
+        max_tokens = int(body.get("max_tokens") or body.get("max_new_tokens") or 5120)
         temperature = float(body.get("temperature") or 1.0)
         top_p = float(body.get("top_p") or 1.0)
 
+        print(
+            f"  [node] processing chat model={model_name!r} stream={stream} "
+            f"max_tokens={max_tokens}{self._request_log_suffix()}",
+            flush=True,
+        )
+
         # Fast path: this node owns the complete model — use HF generate() with KV cache.
         # Avoids the single-token-per-forward-pass limitation of the distributed path.
         if backend.is_head and backend.is_tail:
+            gen_started = time.monotonic()
             try:
                 if stream:
-                    self._stream_openai_response(
+                    token_count = 0
-                        backend.generate_text_streaming(messages, max_tokens, temperature, top_p),
+
-                        model_name,
+                    def _counting_stream():
+                        nonlocal token_count
+                        for token_text in backend.generate_text_streaming(
+                            messages, max_tokens, temperature, top_p,
+                        ):
+                            if token_text:
+                                token_count += 1
+                            yield token_text
+
+                    self._stream_openai_response(_counting_stream(), model_name)
+                    print(
+                        f"  [node] chat complete (stream) tokens={token_count} "
+                        f"elapsed_s={time.monotonic() - gen_started:.1f}{self._request_log_suffix()}",
+                        flush=True,
                     )
                 else:
                     text = backend.generate_text(messages, max_tokens, temperature, top_p)
+                    completion_tokens = _backend_token_count(
+                        backend, "count_text_tokens", text, fallback=len(text.split()) or 1,
+                    )
+                    print(
+                        f"  [node] chat complete tokens={completion_tokens} "
+                        f"elapsed_s={time.monotonic() - gen_started:.1f}{self._request_log_suffix()}",
+                        flush=True,
+                    )
                     self._send_openai_response(text, model_name, False, messages, backend=backend)
             except Exception as exc:
                 self._record_failed_request()
+                print(
+                    f"  [node] chat failed after {time.monotonic() - gen_started:.1f}s: {exc}"
+                    f"{self._request_log_suffix()}",
+                    flush=True,
+                )
                 self._send_json(500, {"error": f"generation failed: {exc}"})
             return
 
@@ -368,7 +415,11 @@ class _TorchHandler(http.server.BaseHTTPRequestHandler):
         if stream:
             stream_emit = self._start_openai_stream(model_name)
 
-        for _ in range(max_tokens):
+        _GENERATION_LOG_INTERVAL = 5.0
+        gen_started = time.monotonic()
+        last_gen_log = gen_started
+
+        for step in range(max_tokens):
             try:
                 payload = backend.encode_prompt(current_text)
             except Exception as exc:
@@ -386,6 +437,21 @@ class _TorchHandler(http.server.BaseHTTPRequestHandler):
             if stream_emit is not None:
                 stream_emit(token_str)
             current_text = current_text + token_str
+            now = time.monotonic()
+            if step == 0 or now - last_gen_log >= _GENERATION_LOG_INTERVAL:
+                print(
+                    f"  [node] generating step={step + 1}/{max_tokens} "
+                    f"tokens={len(generated)} elapsed_s={now - gen_started:.1f}",
+                    flush=True,
+                )
+                last_gen_log = now
+
+        if generated:
+            print(
+                f"  [node] generation complete tokens={len(generated)} "
+                f"elapsed_s={time.monotonic() - gen_started:.1f}",
+                flush=True,
+            )
 
         result_text = "".join(generated)
         if stream_emit is not None:

packages/tracker/meshnet_tracker/dashboard.html

@@ -5,17 +5,24 @@
 <meta name="viewport" content="width=device-width, initial-scale=1">
 <title>meshnet tracker</title>
 <style>
-  :root { --bg:#0d1117; --panel:#161b22; --border:#30363d; --fg:#c9d1d9;
+  :root { --bg:#0d1117; --panel:#161b22; --border:#30363d; --fg:#e6edf3;
-          --dim:#8b949e; --accent:#58a6ff; --ok:#3fb950; --bad:#f85149; --warn:#d29922; }
+          --dim:#8b949e; --accent:#58a6ff; --ok:#3fb950; --bad:#f85149; --warn:#d29922;
+          --chat-input-bg:#21262d; --chat-user-bg:#1f4788; --chat-user-border:#388bfd; }
   * { box-sizing:border-box; }
+  html, body { height:100%; }
   body { margin:0; background:var(--bg); color:var(--fg);
          font:13px/1.5 ui-monospace,SFMono-Regular,Menlo,Consolas,monospace; }
+  body.chat-tab-active { overflow:hidden; height:100dvh; display:flex; flex-direction:column; }
   header { display:flex; align-items:baseline; gap:14px; padding:14px 20px;
-           border-bottom:1px solid var(--border); }
+           border-bottom:1px solid var(--border); flex-shrink:0; }
   header h1 { font-size:16px; margin:0; color:var(--accent); }
   header .meta { color:var(--dim); font-size:12px; }
   main { display:grid; grid-template-columns:repeat(auto-fit,minmax(340px,1fr));
          gap:14px; padding:14px 20px; }
+  body.chat-tab-active main {
+    flex:1; min-height:0; display:flex; flex-direction:column;
+    padding:0; gap:0; overflow:hidden;
+  }
   section { background:var(--panel); border:1px solid var(--border);
             border-radius:8px; padding:12px 14px; min-height:80px; }
   section h2 { margin:0 0 8px; font-size:12px; text-transform:uppercase;
@@ -53,27 +60,132 @@
   .tabs { display:flex; gap:10px; margin-bottom:8px; }
   .tabs a { color:var(--dim); cursor:pointer; }
   .tabs a.active { color:var(--accent); border-bottom:1px solid var(--accent); }
-  .dashboard-tabs { display:flex; gap:10px; padding:10px 20px 0; border-bottom:1px solid var(--border); }
+  .dashboard-tabs { display:flex; gap:10px; padding:10px 20px 0; border-bottom:1px solid var(--border); flex-shrink:0; }
   .dashboard-tabs button { border:0; border-bottom:1px solid transparent; border-radius:0;
           background:transparent; color:var(--dim); padding:5px 0 8px; }
   .dashboard-tabs button.active { color:var(--accent); border-bottom-color:var(--accent); }
   .wide { grid-column:1 / -1; }
   section[hidden] { display:none !important; }
-  .chat-shell { display:grid; grid-template-columns:minmax(0, 1.35fr) minmax(320px, 0.65fr); gap:12px; }
+  section.chat-section {
-  .chat-pane { display:flex; flex-direction:column; gap:10px; min-width:0; }
+    padding:0; border:0; border-radius:0; background:var(--bg); min-height:0;
-  .chat-panel { background:var(--bg); border:1px solid var(--border); border-radius:6px; padding:10px; }
+  }
-  .chat-controls { display:flex; gap:10px; align-items:end; flex-wrap:wrap; }
+  body.chat-tab-active section.chat-section {
-  .chat-controls label { display:flex; flex-direction:column; gap:4px; color:var(--dim); }
+    flex:1; display:flex !important; flex-direction:column; min-height:0;
-  .chat-controls select { min-width:220px; }
+  }
-  .chat-history { display:flex; flex-direction:column; gap:8px; min-height:220px; max-height:420px; overflow:auto; }
+  .chat-app {
-  .chat-message { border:1px solid #21262d; border-radius:6px; padding:8px 10px; background:#10151d; }
+    display:grid; grid-template-columns:260px minmax(0, 1fr); gap:0;
-  .chat-role { color:var(--dim); font-size:11px; text-transform:uppercase; letter-spacing:.06em; margin-bottom:4px; }
+    flex:1; min-height:0; overflow:hidden; background:var(--bg);
-  .chat-role-user { color:var(--accent); }
+  }
-  .chat-role-assistant { color:var(--ok); }
+  .chat-sidebar {
-  .chat-role-error { color:var(--bad); }
+    display:flex; flex-direction:column; min-height:0;
-  .chat-compose { display:flex; flex-direction:column; gap:8px; }
+    border-right:1px solid var(--border); background:var(--panel);
-  .chat-compose textarea { min-height:112px; resize:vertical; width:100%; }
+  }
-  .chat-status { color:var(--dim); font-size:12px; }
+  .chat-new-btn {
+    margin:12px; width:calc(100% - 24px); text-align:left;
+    border:1px solid var(--border); border-radius:8px; padding:10px 12px;
+    background:transparent; color:var(--fg);
+  }
+  .chat-new-btn:hover { background:#10151d; border-color:var(--accent); }
+  .chat-session-list {
+    flex:1; overflow:auto; padding:0 8px 12px; display:flex; flex-direction:column; gap:2px;
+  }
+  .chat-session-list.empty-state {
+    justify-content:center; align-items:center; color:var(--dim); font-style:italic;
+    padding:24px 12px;
+  }
+  .chat-session-item {
+    position:relative; display:block; width:100%; text-align:left;
+    padding:10px 32px 10px 12px; border:1px solid transparent; border-radius:8px;
+    background:transparent; color:var(--fg); cursor:pointer;
+  }
+  .chat-session-item:hover { background:#10151d; }
+  .chat-session-item.active { background:#10151d; border-color:#30363d; }
+  .chat-session-title {
+    font-size:13px; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;
+  }
+  .chat-session-meta {
+    margin-top:3px; font-size:11px; color:var(--dim);
+    white-space:nowrap; overflow:hidden; text-overflow:ellipsis;
+  }
+  .chat-session-delete {
+    position:absolute; top:50%; right:6px; transform:translateY(-50%);
+    padding:2px 6px; min-width:0; border:0; border-radius:4px;
+    background:transparent; color:var(--dim); line-height:1.2; opacity:0;
+  }
+  .chat-session-item:hover .chat-session-delete,
+  .chat-session-item.active .chat-session-delete { opacity:1; }
+  .chat-session-delete:hover { color:var(--bad); background:#1a1012; }
+  .chat-main { display:flex; flex-direction:column; min-height:0; min-width:0; color-scheme:dark; }
+  .chat-toolbar {
+    display:flex; gap:12px; align-items:center; flex-shrink:0;
+    padding:10px 16px; border-bottom:1px solid var(--border); background:var(--panel);
+  }
+  .chat-toolbar label {
+    display:flex; align-items:center; gap:8px; color:var(--dim); margin:0;
+  }
+  .chat-toolbar select {
+    min-width:220px; max-width:min(420px, 50vw);
+    color:var(--fg); background:var(--chat-input-bg); border:1px solid var(--border);
+    border-radius:6px; padding:6px 8px;
+  }
+  .chat-status { color:var(--dim); font-size:12px; margin-left:auto; }
+  .chat-messages {
+    flex:1; overflow:auto; padding:24px 16px; min-height:0;
+    background:var(--bg); color:var(--fg);
+  }
+  .chat-messages-inner {
+    max-width:768px; margin:0 auto; display:flex; flex-direction:column; gap:20px;
+  }
+  .chat-messages.empty .chat-messages-inner {
+    min-height:100%; justify-content:center; align-items:center;
+    color:var(--dim); font-size:14px;
+  }
+  .chat-row { display:flex; width:100%; }
+  .chat-row.user { justify-content:flex-end; }
+  .chat-row.assistant, .chat-row.error { justify-content:flex-start; }
+  .chat-bubble {
+    max-width:85%; padding:12px 14px; border-radius:16px; line-height:1.55;
+    white-space:pre-wrap; word-break:break-word; font-size:14px; color:var(--fg);
+  }
+  .chat-bubble.user {
+    background:var(--chat-user-bg); border:1px solid var(--chat-user-border);
+    border-bottom-right-radius:4px; color:#f0f6fc;
+  }
+  .chat-bubble.assistant {
+    background:var(--panel); border:1px solid var(--border);
+    border-bottom-left-radius:4px; max-width:100%; color:var(--fg);
+  }
+  .chat-bubble.error {
+    background:#1a1012; border:1px solid #5c2020; color:#ffb4b4; border-bottom-left-radius:4px;
+  }
+  .chat-compose-wrap {
+    flex-shrink:0; padding:12px 16px 16px; border-top:1px solid var(--border);
+    background:var(--panel);
+  }
+  .chat-compose {
+    display:flex; gap:8px; align-items:flex-end; max-width:768px; margin:0 auto;
+    padding:10px 12px; border:1px solid var(--border); border-radius:16px;
+    background:var(--chat-input-bg);
+  }
+  .chat-compose:focus-within {
+    border-color:var(--accent);
+    box-shadow:0 0 0 1px var(--accent);
+  }
+  .chat-compose textarea {
+    flex:1; min-height:24px; max-height:200px; resize:none; width:auto;
+    border:0; background:transparent; padding:4px 0; outline:none;
+    color:var(--fg); caret-color:var(--accent); font:inherit; font-size:14px; line-height:1.5;
+  }
+  .chat-compose textarea::placeholder { color:var(--dim); opacity:1; }
+  .chat-compose button {
+    flex-shrink:0; min-width:36px; height:36px; padding:0;
+    border-radius:8px; border:1px solid var(--chat-user-border);
+    background:var(--chat-user-bg); color:#f0f6fc;
+  }
+  .chat-compose button:hover:not(:disabled) {
+    border-color:var(--accent); background:#2563b8;
+  }
+  .chat-compose button:disabled { opacity:.45; cursor:not-allowed; }
   .console {
     background:var(--bg); border:1px solid var(--border); border-radius:6px;
     min-height:160px; max-height:280px; overflow:auto; padding:7px 9px;
@@ -88,6 +200,9 @@
   .status-processing { color:var(--accent); }
   .status-failed { color:var(--bad); }
   .status-complete { color:var(--ok); }
+  .status-canceled { color:var(--dim); }
+  button.btn-cancel { color:var(--dim); padding:0 5px; min-width:1.4em; line-height:1.2; }
+  button.btn-cancel:hover { color:var(--bad); border-color:var(--bad); }
 </style>
 </head>
 <body>
@@ -108,27 +223,27 @@
   <section data-tab="overview"><h2>Nodes &amp; coverage</h2><div id="nodes" class="empty">loading…</div></section>
   <section data-tab="overview"><h2>Model usage (RPM)</h2><div id="stats" class="empty">loading…</div></section>
   <section data-tab="overview" class="wide"><h2>Call wall</h2><div id="call-wall" class="empty">loading...</div></section>
-  <section data-tab="chat" class="wide">
+  <section data-tab="chat" class="wide chat-section">
-    <h2>Chat / inference</h2>
+    <div class="chat-app">
-    <div class="chat-shell">
+      <aside class="chat-sidebar">
-      <div class="chat-pane">
+        <button type="button" class="chat-new-btn" onclick="createNewChatSession()">+ New chat</button>
-        <div class="chat-panel chat-controls">
+        <div id="chat-session-list" class="chat-session-list empty-state">No chats yet</div>
+      </aside>
+      <div class="chat-main">
+        <div class="chat-toolbar">
           <label>Model
             <select id="chat-model" onchange="selectChatModel(this.value)"></select>
           </label>
-          <button class="small" onclick="clearChatHistory()">clear history</button>
-        </div>
-        <div class="chat-panel chat-compose">
-          <textarea id="chat-prompt" placeholder="Ask a question or describe the task"></textarea>
-          <div class="form-row">
-            <button onclick="sendChat()" id="chat-send">Send</button>
-          </div>
-        </div>
-      </div>
-      <div class="chat-pane">
-        <div class="chat-panel">
           <div id="chat-status" class="chat-status">select a model to start</div>
-          <div id="chat-history" class="chat-history empty">no messages yet</div>
+        </div>
+        <div id="chat-history" class="chat-messages empty">
+          <div class="chat-messages-inner">Send a message to start this conversation.</div>
+        </div>
+        <div class="chat-compose-wrap">
+          <div class="chat-compose">
+            <textarea id="chat-prompt" placeholder="Message…" rows="1" aria-label="Message"></textarea>
+            <button type="button" onclick="sendChat()" id="chat-send" title="Send (Enter)">↑</button>
+          </div>
         </div>
       </div>
     </div>
@@ -350,6 +465,13 @@ function buildCallWallStates(events) {
       rec.elapsed = f.elapsed_seconds;
       rec.stream = f.stream;
       rec.terminal = e;
+    } else if (msg === "proxy canceled") {
+      rec.status = "canceled";
+      rec.model = rec.model || f.model || f.route_model || "?";
+      rec.tokens = f.tokens;
+      rec.tps = f.tokens_per_sec;
+      rec.elapsed = f.elapsed_seconds;
+      rec.terminal = e;
     } else if (msg === "proxy failed" || msg === "direct proxy failed after relay") {
       rec.status = "failed";
       rec.model = rec.model || f.model || f.route_model || "?";
@@ -379,7 +501,7 @@ function renderCallWall(consoleData, stats) {
   const terminal = [];
   for (const rec of states.values()) {
     if (rec.status === "pending" || rec.status === "processing") active.push(rec);
-    else if (rec.status === "complete" || rec.status === "failed") terminal.push(rec);
+    else if (rec.status === "complete" || rec.status === "failed" || rec.status === "canceled") terminal.push(rec);
   }
   active.sort((a, b) => (a.started || 0) - (b.started || 0));
   terminal.sort((a, b) => (b.terminal && b.terminal.ts) - (a.terminal && a.terminal.ts));
@@ -401,10 +523,13 @@ function renderCallWall(consoleData, stats) {
     `</div>`;
 
   if (active.length) {
-    html += table(["status", "age", "model", "request", "live tps", "tokens", "queue", "route / note"], active.map(rec => {
+    const canCancelProxies = isAdmin || !isLoggedIn;
+    const headers = ["status", "age", "model", "request", "live tps", "tokens", "queue", "route / note"];
+    if (canCancelProxies) headers.push("");
+    html += table(headers, active.map(rec => {
       const statusCls = rec.status === "pending" ? "status-pending" : "status-processing";
       const note = rec.warn || (rec.route ? short(String(rec.route), 28) : "");
-      return [
+      const row = [
         `<span class="${statusCls}">${esc(rec.status)}</span>`,
         `<span class="num">${esc(callWallAgeSeconds(rec, nowSec).toFixed(1))}s</span>`,
         esc(short(rec.model || "?", 28)),
@@ -414,6 +539,12 @@ function renderCallWall(consoleData, stats) {
         `<span class="num">${esc(String(callWallMaxQueue(rec)))}</span>`,
         esc(note),
       ];
+      if (canCancelProxies) {
+        row.push(
+          `<button type="button" class="small btn-cancel" data-cancel-request="${esc(rec.id)}" title="Cancel request">×</button>`,
+        );
+      }
+      return row;
     }));
   } else {
     html += '<div class="empty">no in-flight requests</div>';
@@ -422,10 +553,16 @@ function renderCallWall(consoleData, stats) {
   const historyRows = terminal.slice(0, 40).map(rec => {
     const e = rec.terminal || {};
     const f = e.fields || {};
-    const statusCls = rec.status === "failed" ? "status-failed" : "status-complete";
+    const statusCls = rec.status === "failed"
+      ? "status-failed"
+      : rec.status === "canceled"
+        ? "status-canceled"
+        : "status-complete";
     const detail = rec.status === "failed"
       ? esc(short(rec.error || "?", 40))
-      : (f.stream ? "stream" : "json");
+      : rec.status === "canceled"
+        ? "canceled"
+        : (f.stream ? "stream" : "json");
     return [
       new Date((e.ts || 0) * 1000).toLocaleTimeString(),
       `<span class="${statusCls}">${esc(rec.status)}</span>`,
@@ -437,7 +574,7 @@ function renderCallWall(consoleData, stats) {
       detail,
     ];
   });
-  html += '<div style="margin-top:8px"><b class="dim">recent completed / failed</b></div>';
+  html += '<div style="margin-top:8px"><b class="dim">recent completed / failed / canceled</b></div>';
   html += historyRows.length
     ? table(["time", "status", "model", "request", "tps", "tokens", "sec", "detail"], historyRows)
     : '<div class="empty">no completed requests yet</div>';
@@ -579,17 +716,214 @@ let lastStats = null;
 let availableModels = [];
 let chatHistory = [];
 let chatBusy = false;
+let chatSessions = [];
+let activeChatSessionId = "";
 let selectedChatModel = localStorage.getItem("meshnet_chat_model") || "";
+const CHAT_SESSIONS_KEY = "meshnet_chat_sessions_v1";
+const CHAT_ACTIVE_SESSION_KEY = "meshnet_chat_active_session_v1";
+const CHAT_SESSIONS_LIMIT = 50;
+
+function newChatSessionId() {
+  if (window.crypto && crypto.randomUUID) return crypto.randomUUID();
+  return "chat-" + Date.now().toString(36) + "-" + Math.random().toString(36).slice(2, 8);
+}
+
+function loadChatSessionsStore() {
+  try {
+    const raw = localStorage.getItem(CHAT_SESSIONS_KEY);
+    const parsed = raw ? JSON.parse(raw) : [];
+    return Array.isArray(parsed) ? parsed : [];
+  } catch {
+    return [];
+  }
+}
+
+function saveChatSessionsStore() {
+  localStorage.setItem(CHAT_SESSIONS_KEY, JSON.stringify(chatSessions));
+  if (activeChatSessionId) {
+    localStorage.setItem(CHAT_ACTIVE_SESSION_KEY, activeChatSessionId);
+  }
+}
+
+function chatSessionTitle(session) {
+  const firstUser = (session.messages || []).find(msg => msg.role === "user");
+  if (!firstUser || !firstUser.content) return "New chat";
+  const text = String(firstUser.content).trim().replace(/\s+/g, " ");
+  return text.length > 42 ? text.slice(0, 42) + "…" : text;
+}
+
+function formatSessionTime(iso) {
+  if (!iso) return "";
+  const date = new Date(iso);
+  if (Number.isNaN(date.getTime())) return "";
+  const now = new Date();
+  const sameDay = date.toDateString() === now.toDateString();
+  if (sameDay) return date.toLocaleTimeString([], { hour: "2-digit", minute: "2-digit" });
+  return date.toLocaleDateString([], { month: "short", day: "numeric" });
+}
+
+function getActiveChatSession() {
+  return chatSessions.find(session => session.id === activeChatSessionId) || null;
+}
+
+function persistActiveChatSession() {
+  const session = getActiveChatSession();
+  if (!session) return;
+  session.messages = chatHistory.slice();
+  session.model = selectedChatModel || session.model || "";
+  session.title = chatSessionTitle(session);
+  session.updatedAt = new Date().toISOString();
+  chatSessions.sort((a, b) => String(b.updatedAt).localeCompare(String(a.updatedAt)));
+  if (chatSessions.length > CHAT_SESSIONS_LIMIT) {
+    chatSessions = chatSessions.slice(0, CHAT_SESSIONS_LIMIT);
+    if (!chatSessions.some(item => item.id === activeChatSessionId)) {
+      activeChatSessionId = chatSessions[0].id;
+      chatHistory = chatSessions[0].messages.slice();
+    }
+  }
+  saveChatSessionsStore();
+  renderChatSessionList();
+}
+
+function clearChatPrompt() {
+  const promptEl = $("chat-prompt");
+  if (!promptEl) return;
+  promptEl.value = "";
+  promptEl.style.height = "auto";
+}
+
+function createNewChatSession() {
+  if (chatBusy) return;
+  const session = {
+    id: newChatSessionId(),
+    title: "New chat",
+    model: selectedChatModel || "",
+    messages: [],
+    createdAt: new Date().toISOString(),
+    updatedAt: new Date().toISOString(),
+  };
+  chatSessions.unshift(session);
+  activeChatSessionId = session.id;
+  chatHistory = [];
+  clearChatPrompt();
+  saveChatSessionsStore();
+  renderChatSessionList();
+  renderChatHistory();
+  renderChatAuthHint();
+  const promptEl = $("chat-prompt");
+  if (promptEl) promptEl.focus();
+}
+
+function selectChatSession(sessionId) {
+  if (chatBusy) return;
+  const session = chatSessions.find(item => item.id === sessionId);
+  if (!session) return;
+  if (sessionId === activeChatSessionId) return;
+  activeChatSessionId = session.id;
+  chatHistory = (session.messages || []).slice();
+  clearChatPrompt();
+  if (session.model) {
+    selectedChatModel = session.model;
+    localStorage.setItem("meshnet_chat_model", selectedChatModel);
+    const select = $("chat-model");
+    if (select) select.value = selectedChatModel;
+  }
+  localStorage.setItem(CHAT_ACTIVE_SESSION_KEY, activeChatSessionId);
+  renderChatSessionList();
+  renderChatHistory();
+  renderChatAuthHint();
+}
+
+function deleteChatSession(sessionId, event) {
+  if (event) {
+    event.preventDefault();
+    event.stopPropagation();
+  }
+  if (chatBusy) return;
+  const index = chatSessions.findIndex(item => item.id === sessionId);
+  if (index < 0) return;
+  chatSessions.splice(index, 1);
+  if (activeChatSessionId === sessionId) {
+    if (chatSessions.length) {
+      activeChatSessionId = chatSessions[0].id;
+      chatHistory = (chatSessions[0].messages || []).slice();
+      clearChatPrompt();
+      if (chatSessions[0].model) {
+        selectedChatModel = chatSessions[0].model;
+        localStorage.setItem("meshnet_chat_model", selectedChatModel);
+      }
+    } else {
+      saveChatSessionsStore();
+      createNewChatSession();
+      return;
+    }
+  }
+  saveChatSessionsStore();
+  renderChatSessionList();
+  renderChatHistory();
+  renderChatModels();
+}
+
+function initChatSessions() {
+  chatSessions = loadChatSessionsStore();
+  activeChatSessionId = localStorage.getItem(CHAT_ACTIVE_SESSION_KEY) || "";
+  const active = chatSessions.find(session => session.id === activeChatSessionId);
+  if (!active) {
+    if (chatSessions.length) {
+      activeChatSessionId = chatSessions[0].id;
+      chatHistory = (chatSessions[0].messages || []).slice();
+      if (chatSessions[0].model) selectedChatModel = chatSessions[0].model;
+    } else {
+      createNewChatSession();
+      return;
+    }
+  } else {
+    chatHistory = (active.messages || []).slice();
+    if (active.model) selectedChatModel = active.model;
+  }
+  renderChatSessionList();
+  renderChatHistory();
+}
+
+function renderChatSessionList() {
+  const list = $("chat-session-list");
+  if (!list) return;
+  if (!chatSessions.length) {
+    list.className = "chat-session-list empty-state";
+    list.innerHTML = "No chats yet";
+    return;
+  }
+  list.className = "chat-session-list";
+  list.innerHTML = chatSessions.map(session => {
+    const active = session.id === activeChatSessionId ? " active" : "";
+    const title = esc(chatSessionTitle(session));
+    const when = esc(formatSessionTime(session.updatedAt || session.createdAt));
+    const id = JSON.stringify(session.id);
+    return `<div class="chat-session-item${active}" role="button" tabindex="0"` +
+      ` onclick="selectChatSession(${id})"` +
+      ` onkeydown="if(event.key==='Enter'||event.key===' '){event.preventDefault();selectChatSession(${id});}">` +
+      `<div class="chat-session-title">${title}</div>` +
+      (when ? `<div class="chat-session-meta">${when}</div>` : "") +
+      `<button type="button" class="chat-session-delete" title="Delete chat"` +
+      ` onclick="deleteChatSession(${id}, event)">×</button>` +
+      `</div>`;
+  }).join("");
+}
 
 function switchDashboardTab(name) {
   if (name === "admin" && !isAdmin) name = "overview";
   if (name === "billing" && !isLoggedIn) name = "overview";
   dashboardTab = name;
+  document.body.classList.toggle("chat-tab-active", name === "chat");
   updateSectionVisibility();
   for (const tabName of ["overview", "chat", "billing", "admin"]) {
     const button = $("tab-" + tabName);
     if (button) button.classList.toggle("active", tabName === dashboardTab);
   }
+  if (name === "chat") {
+    const promptEl = $("chat-prompt");
+    if (promptEl) promptEl.focus();
+  }
 }
 
 function updateSectionVisibility() {
@@ -607,18 +941,18 @@ function renderChatStatus(text) {
 
 function renderChatHistory() {
   const history = $("chat-history");
+  if (!history) return;
   if (!chatHistory.length) {
-    history.classList.add("empty");
+    history.className = "chat-messages empty";
-    history.innerHTML = "no messages yet";
+    history.innerHTML = '<div class="chat-messages-inner">Send a message to start this conversation.</div>';
     return;
   }
-  history.classList.remove("empty");
+  history.className = "chat-messages";
-  history.innerHTML = chatHistory.map(msg => {
+  const rows = chatHistory.map(msg => {
-    const roleClass = msg.role === "user" ? "chat-role-user" : msg.role === "assistant" ? "chat-role-assistant" : "chat-role-error";
+    const role = msg.role === "user" ? "user" : msg.role === "assistant" ? "assistant" : "error";
-    const label = msg.role === "user" ? "user" : msg.role === "assistant" ? "assistant" : "error";
+    return `<div class="chat-row ${role}"><div class="chat-bubble ${role}">${esc(msg.content)}</div></div>`;
-    const meta = msg.model ? ` <span class="dim">· ${esc(short(msg.model, 24))}</span>` : "";
-    return `<div class="chat-message"><div class="chat-role ${roleClass}">${label}${meta}</div><div>${esc(msg.content)}</div></div>`;
   }).join("");
+  history.innerHTML = `<div class="chat-messages-inner">${rows}</div>`;
   history.scrollTop = history.scrollHeight;
 }
 
@@ -649,12 +983,13 @@ function renderChatModels() {
 function selectChatModel(value) {
   selectedChatModel = value || "";
   localStorage.setItem("meshnet_chat_model", selectedChatModel);
-}
+  const session = getActiveChatSession();
-
+  if (session) {
-function clearChatHistory() {
+    session.model = selectedChatModel;
-  chatHistory = [];
+    session.updatedAt = new Date().toISOString();
-  renderChatHistory();
+    saveChatSessionsStore();
-  renderChatStatus("history cleared");
+    renderChatSessionList();
+  }
 }
 
 function chatAuthToken() {
@@ -905,13 +1240,15 @@ async function sendChat() {
       { role: "user", content: prompt },
     ],
     stream: false,
-    max_tokens: 256,
+    max_tokens: 15120,
   };
   chatBusy = true;
   $("chat-send").disabled = true;
   promptEl.value = "";
+  promptEl.style.height = "auto";
   chatHistory.push({ role: "user", content: prompt, model: selectedChatModel });
   renderChatHistory();
+  persistActiveChatSession();
   renderChatStatus("sending request…");
   const r = await apiCall("/v1/chat/completions", "POST", body, bearerToken);
   chatBusy = false;
@@ -922,6 +1259,7 @@ async function sendChat() {
       : "request failed";
     chatHistory.push({ role: "error", content: error, model: selectedChatModel });
     renderChatHistory();
+    persistActiveChatSession();
     renderChatStatus(error);
     promptEl.focus();
     return;
@@ -934,12 +1272,29 @@ async function sendChat() {
     model: selectedChatModel,
   });
   renderChatHistory();
+  persistActiveChatSession();
   renderChatStatus(usage
     ? `done: ${usage.total_tokens ?? "?"} tokens`
     : "done");
   promptEl.focus();
 }
 
+function bindChatPromptShortcuts() {
+  const promptEl = $("chat-prompt");
+  if (!promptEl || promptEl.dataset.bound === "1") return;
+  promptEl.dataset.bound = "1";
+  promptEl.addEventListener("keydown", event => {
+    if (event.key === "Enter" && !event.shiftKey) {
+      event.preventDefault();
+      sendChat();
+    }
+  });
+  promptEl.addEventListener("input", () => {
+    promptEl.style.height = "auto";
+    promptEl.style.height = Math.min(promptEl.scrollHeight, 200) + "px";
+  });
+}
+
 async function renderAdminPanel() {
   const r = await apiCall("/v1/admin/accounts");
   if (!r.ok) { setAdminMode(false); return; }
@@ -956,6 +1311,23 @@ async function renderAdminPanel() {
   $("admin").innerHTML = table(["account", "role", "keys", "balance (USDT)", "created"], rows);
 }
 
+async function cancelProxyRequest(requestId) {
+  const r = await apiCall(
+    `/v1/proxy/requests/${encodeURIComponent(requestId)}/cancel`,
+    "POST",
+    {},
+  );
+  if (r.ok) refresh();
+}
+
+$("call-wall").addEventListener("click", (event) => {
+  const button = event.target.closest("[data-cancel-request]");
+  if (!button) return;
+  event.preventDefault();
+  const requestId = button.getAttribute("data-cancel-request");
+  if (requestId) cancelProxyRequest(requestId);
+});
+
 async function refresh() {
   $("self-url").textContent = location.host;
   const [raft, map, stats, models, consoleData, adminData] = await Promise.all([
@@ -992,6 +1364,8 @@ async function refresh() {
   $("refreshed").textContent = "refreshed " + new Date().toLocaleTimeString();
 }
 refresh();
+initChatSessions();
+bindChatPromptShortcuts();
 renderAccountPanel();
 renderChatModels();
 renderChatHistory();

packages/tracker/meshnet_tracker/server.py

@@ -35,6 +35,7 @@ import urllib.parse
 import urllib.request
 import uuid
 from collections import deque
+from dataclasses import dataclass, field
 from importlib.resources import files
 from pathlib import Path
 from typing import Any
@@ -51,6 +52,7 @@ from .raft import RaftNode
 
 
 _CONSOLE_LIMIT = 300
+_PROXY_PROGRESS_LOG_INTERVAL = 5.0
 
 
 def _preset_price_keys(name: str, preset: dict) -> set[str]:
@@ -1413,6 +1415,10 @@ def _relay_http_request_frames(
     headers: dict[str, str],
     timeout: float = 310.0,
     idle_timeout: float = 120.0,
+    *,
+    cancel_event: threading.Event | None = None,
+    ws_holder: list[Any] | None = None,
+    ws_lock: threading.Lock | None = None,
 ):
     """Send an HTTP-shaped request through a relay RPC WebSocket, yielding
     response frames until a terminal one (US-036).
@@ -1430,6 +1436,14 @@ def _relay_http_request_frames(
     deadline = time.monotonic() + timeout
     try:
         with wsc.connect(relay_addr, open_timeout=10, close_timeout=5) as ws:
+            if ws_holder is not None:
+                if ws_lock is not None:
+                    with ws_lock:
+                        ws_holder.clear()
+                        ws_holder.append(ws)
+                else:
+                    ws_holder.clear()
+                    ws_holder.append(ws)
             ws.send(json.dumps({
                 "request_id": request_id,
                 "method": "POST",
@@ -1438,6 +1452,8 @@ def _relay_http_request_frames(
                 "body": body.decode(errors="replace"),
             }))
             while True:
+                if cancel_event is not None and cancel_event.is_set():
+                    return
                 remaining = deadline - time.monotonic()
                 if remaining <= 0:
                     return
@@ -2076,7 +2092,15 @@ def _registration_ban_error(contracts: Any | None, wallet_address: str | None) -
     return None
 
 
-def _tracker_log(server: "_TrackerHTTPServer", level: str, message: str, **fields: Any) -> None:
+def _tracker_log(
+    server: "_TrackerHTTPServer",
+    level: str,
+    message: str,
+    *,
+    stdout: bool = True,
+    update_console_key: str | None = None,
+    **fields: Any,
+) -> None:
     event = {
         "ts": time.time(),
         "level": level,
@@ -2088,10 +2112,88 @@ def _tracker_log(server: "_TrackerHTTPServer", level: str, message: str, **field
         },
     }
     with server.console_lock:
-        server.console_events.append(event)
+        if update_console_key is not None:
-    extras = " ".join(f"{key}={value}" for key, value in event["fields"].items())
+            updated = False
-    suffix = f"  {extras}" if extras else ""
+            for existing in reversed(server.console_events):
-    print(f"[tracker] {level}: {message}{suffix}", flush=True)
+                if (
+                    existing.get("message") == message
+                    and existing.get("fields", {}).get("request_id") == update_console_key
+                ):
+                    existing["ts"] = event["ts"]
+                    existing["fields"] = event["fields"]
+                    updated = True
+                    break
+            if not updated:
+                server.console_events.append(event)
+        else:
+            server.console_events.append(event)
+    if stdout:
+        extras = " ".join(f"{key}={value}" for key, value in event["fields"].items())
+        suffix = f"  {extras}" if extras else ""
+        print(f"[tracker] {level}: {message}{suffix}", flush=True)
+
+
+@dataclass
+class _ActiveProxyContext:
+    request_id: str
+    cancel_event: threading.Event = field(default_factory=threading.Event)
+    upstream: Any | None = None
+    upstream_lock: threading.Lock = field(default_factory=threading.Lock)
+    relay_ws: Any | None = None
+    relay_ws_lock: threading.Lock = field(default_factory=threading.Lock)
+
+
+def _register_active_proxy(server: "_TrackerHTTPServer", request_id: str) -> _ActiveProxyContext:
+    ctx = _ActiveProxyContext(request_id=request_id)
+    with server.active_proxies_lock:
+        server.active_proxies[request_id] = ctx
+    return ctx
+
+
+def _unregister_active_proxy(server: "_TrackerHTTPServer", request_id: str) -> None:
+    with server.active_proxies_lock:
+        server.active_proxies.pop(request_id, None)
+
+
+def _request_proxy_cancel(server: "_TrackerHTTPServer", request_id: str) -> bool:
+    with server.active_proxies_lock:
+        ctx = server.active_proxies.get(request_id)
+    if ctx is None:
+        return False
+    ctx.cancel_event.set()
+
+    def _close_resources() -> None:
+        with ctx.upstream_lock:
+            upstream = ctx.upstream
+        if upstream is not None:
+            try:
+                upstream.close()
+            except Exception:
+                pass
+        with ctx.relay_ws_lock:
+            relay_ws = ctx.relay_ws
+        if relay_ws is not None:
+            try:
+                relay_ws.close()
+            except Exception:
+                pass
+
+    threading.Thread(target=_close_resources, daemon=True).start()
+    return True
+
+
+def _set_upstream_read_timeout(upstream: Any, timeout: float) -> None:
+    fp = getattr(upstream, "fp", None)
+    raw = getattr(fp, "raw", None) if fp is not None else None
+    sock = getattr(raw, "_sock", None) if raw is not None else None
+    if sock is not None:
+        sock.settimeout(timeout)
+
+
+def _clear_proxy_progress_log_state(server: "_TrackerHTTPServer", request_id: str) -> None:
+    state = getattr(server, "_proxy_progress_log_state", None)
+    if state is not None:
+        state.pop(request_id, None)
 
 
 def _tracker_log_proxy_progress(
@@ -2108,10 +2210,21 @@ def _tracker_log_proxy_progress(
 ) -> None:
     elapsed = time.monotonic() - started
     effective_elapsed = max(elapsed, 1e-6)
+    now = time.monotonic()
+    state = getattr(server, "_proxy_progress_log_state", None)
+    if state is None:
+        state = {}
+        server._proxy_progress_log_state = state
+    last_stdout = state.get(request_id)
+    stdout = last_stdout is None or (now - last_stdout) >= _PROXY_PROGRESS_LOG_INTERVAL
+    if stdout:
+        state[request_id] = now
     _tracker_log(
         server,
         "info",
         "proxy progress",
+        stdout=stdout,
+        update_console_key=request_id,
         request_id=request_id,
         model=model,
         route_model=route_model,
@@ -2209,6 +2322,8 @@ class _TrackerHTTPServer(socketserver.ThreadingMixIn, http.server.HTTPServer):
         self.models_dir = models_dir
         self.console_events = deque(maxlen=_CONSOLE_LIMIT)
         self.console_lock = threading.Lock()
+        self.active_proxies: dict[str, _ActiveProxyContext] = {}
+        self.active_proxies_lock = threading.Lock()
 
 
 class _TrackerHandler(http.server.BaseHTTPRequestHandler):
@@ -2364,6 +2479,17 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
         if len(parts) == 5 and parts[1] == "v1" and parts[2] == "nodes" and parts[4] == "heartbeat":
             self._handle_heartbeat(parts[3])
             return
+        # /v1/proxy/requests/<request_id>/cancel
+        if (
+            len(parts) == 6
+            and parts[1] == "v1"
+            and parts[2] == "proxy"
+            and parts[3] == "requests"
+            and parts[5] == "cancel"
+            and parts[4]
+        ):
+            self._handle_proxy_request_cancel(urllib.parse.unquote(parts[4]))
+            return
         self.send_response(404)
         self.end_headers()
 
@@ -2886,6 +3012,9 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
             if inflight_recorded:
                 _record_proxy_inflight(server, inflight_nodes, -1)
                 inflight_recorded = False
+            _unregister_active_proxy(server, request_id)
+
+        proxy_ctx = _register_active_proxy(server, request_id)
 
         _tracker_log(
             server,
@@ -2906,6 +3035,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
             headers={
                 "Content-Type": "application/json",
                 "X-Meshnet-Route": downstream_urls,
+                "X-Meshnet-Request-Id": request_id,
             },
             method="POST",
         )
@@ -2917,6 +3047,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
         relay_headers = {
             "Content-Type": "application/json",
             "X-Meshnet-Route": downstream_urls,
+            "X-Meshnet-Request-Id": request_id,
             **({"Authorization": auth} if auth else {}),
         }
 
@@ -2930,13 +3061,34 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                 direct_endpoint=target_url,
             )
             started = time.monotonic()
+            relay_ws_holder: list[Any] = []
             frames = _relay_http_request_frames(
                 node.relay_addr,
                 path="/v1/chat/completions",
                 body=raw_body,
                 headers=relay_headers,
+                cancel_event=proxy_ctx.cancel_event,
+                ws_holder=relay_ws_holder,
+                ws_lock=proxy_ctx.relay_ws_lock,
             )
             first = next(frames, None)
+            with proxy_ctx.relay_ws_lock:
+                proxy_ctx.relay_ws = relay_ws_holder[0] if relay_ws_holder else None
+            if proxy_ctx.cancel_event.is_set():
+                if self._finalize_proxy_cancel(
+                    proxy_ctx=proxy_ctx,
+                    server=server,
+                    request_id=request_id,
+                    started=started,
+                    model=model,
+                    route_model=route_model,
+                    route_nodes=route_nodes,
+                    api_key=api_key,
+                    node_work=node_work,
+                    body=body,
+                    finish_proxy_inflight=finish_proxy_inflight,
+                ):
+                    return
             if first is not None and first.get("stream"):
                 # Streamed response (US-036): forward SSE chunks as they arrive
                 # and run the same token accounting as the direct stream path.
@@ -2945,6 +3097,8 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                     model, route_model, route_nodes, api_key, node_work,
                     request_body=body,
                     request_id=request_id,
+                    proxy_ctx=proxy_ctx,
+                    finish_proxy_inflight=finish_proxy_inflight,
                 )
                 finish_proxy_inflight()
                 return
@@ -2959,6 +3113,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                         in_tokens, out_tokens = 0, 0
                     tokens = in_tokens + out_tokens
                     self._record_observed_throughput(model, route_model, tokens, elapsed, route_nodes)
+                    _clear_proxy_progress_log_state(server, request_id)
                     _tracker_log(
                         server,
                         "info",
@@ -2989,7 +3144,69 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
 
         try:
             started = time.monotonic()
-            upstream = urllib.request.urlopen(req, timeout=300.0)
+            upstream_result: list[Any] = []
+            connect_errors: list[BaseException] = []
+
+            def _connect_upstream() -> None:
+                try:
+                    upstream_result.append(urllib.request.urlopen(req, timeout=300.0))
+                except BaseException as exc:
+                    connect_errors.append(exc)
+
+            connect_thread = threading.Thread(target=_connect_upstream, daemon=True)
+            connect_thread.start()
+            while connect_thread.is_alive():
+                if proxy_ctx.cancel_event.is_set():
+                    connect_thread.join(timeout=310.0)
+                    if upstream_result:
+                        try:
+                            upstream_result[0].close()
+                        except Exception:
+                            pass
+                    if self._finalize_proxy_cancel(
+                        proxy_ctx=proxy_ctx,
+                        server=server,
+                        request_id=request_id,
+                        started=started,
+                        model=model,
+                        route_model=route_model,
+                        route_nodes=route_nodes,
+                        api_key=api_key,
+                        node_work=node_work,
+                        body=body,
+                        finish_proxy_inflight=finish_proxy_inflight,
+                    ):
+                        return
+                connect_thread.join(0.2)
+
+            if proxy_ctx.cancel_event.is_set():
+                if upstream_result:
+                    try:
+                        upstream_result[0].close()
+                    except Exception:
+                        pass
+                if self._finalize_proxy_cancel(
+                    proxy_ctx=proxy_ctx,
+                    server=server,
+                    request_id=request_id,
+                    started=started,
+                    model=model,
+                    route_model=route_model,
+                    route_nodes=route_nodes,
+                    api_key=api_key,
+                    node_work=node_work,
+                    body=body,
+                    finish_proxy_inflight=finish_proxy_inflight,
+                ):
+                    return
+
+            if connect_errors:
+                raise connect_errors[0]
+
+            upstream = upstream_result[0]
+            with proxy_ctx.upstream_lock:
+                proxy_ctx.upstream = upstream
+            _set_upstream_read_timeout(upstream, 0.5)
             _tracker_log(server, "info", "proxy connected", request_id=request_id, target_url=target_url)
         except urllib.error.HTTPError as exc:
             # Relay error status + body from node
@@ -3002,9 +3219,11 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                 self.wfile.write(err_body)
             except BrokenPipeError:
                 pass
+            _clear_proxy_progress_log_state(server, request_id)
             finish_proxy_inflight()
             return
         except Exception as exc:
+            _clear_proxy_progress_log_state(server, request_id)
             if node.relay_addr:
                 _tracker_log(
                     server,
@@ -3045,8 +3264,15 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                 client_gone = False
                 try:
                     while True:
-                        line = upstream.readline()
+                        if proxy_ctx.cancel_event.is_set():
+                            break
+                        try:
+                            line = upstream.readline()
+                        except TimeoutError:
+                            continue
                         if not line:
+                            if proxy_ctx.cancel_event.is_set():
+                                break
                             break
                         if not client_gone:
                             try:
@@ -3071,6 +3297,22 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                             stream_usage = usage
                 except (BrokenPipeError, ConnectionResetError):
                     pass
+                if self._finalize_proxy_cancel(
+                    proxy_ctx=proxy_ctx,
+                    server=server,
+                    request_id=request_id,
+                    started=started,
+                    model=model,
+                    route_model=route_model,
+                    route_nodes=route_nodes,
+                    api_key=api_key,
+                    node_work=node_work,
+                    body=body,
+                    finish_proxy_inflight=finish_proxy_inflight,
+                    observed_stream_tokens=observed_stream_tokens,
+                    stream_usage=stream_usage,
+                ):
+                    return
                 elapsed = time.monotonic() - started
                 # Bill even on client disconnect — the nodes did the work.
                 # Observed stream chunks are authoritative for the upper bound;
@@ -3082,6 +3324,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                     model, route_model, in_tokens + out_tokens, elapsed, route_nodes
                 )
                 tokens = in_tokens + out_tokens
+                _clear_proxy_progress_log_state(server, request_id)
                 _tracker_log(
                     server,
                     "info",
@@ -3113,6 +3356,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                     in_tokens, out_tokens = 0, 0
                 tokens = in_tokens + out_tokens
                 self._record_observed_throughput(model, route_model, tokens, elapsed, route_nodes)
+                _clear_proxy_progress_log_state(server, request_id)
                 _tracker_log(
                     server,
                     "info",
@@ -3272,6 +3516,9 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
         node_work: list,
         request_body: dict,
         request_id: str,
+        *,
+        proxy_ctx: _ActiveProxyContext | None = None,
+        finish_proxy_inflight: Any = None,
     ) -> None:
         """Forward a streamed relay response (US-036) to the client as SSE,
         billing with the same accounting as the direct stream path."""
@@ -3285,6 +3532,8 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
         observed_stream_tokens = 0
         client_gone = False
         for frame in itertools.chain([first], frames):
+            if proxy_ctx is not None and proxy_ctx.cancel_event.is_set():
+                break
             chunk = frame.get("chunk") or ""
             if not chunk:
                 continue
@@ -3312,6 +3561,26 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
                     )
                 if usage is not None:
                     stream_usage = usage
+        if (
+            proxy_ctx is not None
+            and finish_proxy_inflight is not None
+            and self._finalize_proxy_cancel(
+                proxy_ctx=proxy_ctx,
+                server=server,
+                request_id=request_id,
+                started=started,
+                model=model,
+                route_model=route_model,
+                route_nodes=route_nodes,
+                api_key=api_key,
+                node_work=node_work,
+                body=request_body,
+                finish_proxy_inflight=finish_proxy_inflight,
+                observed_stream_tokens=observed_stream_tokens,
+                stream_usage=stream_usage,
+            )
+        ):
+            return
         elapsed = time.monotonic() - started
         in_tokens, out_tokens = _stream_billable_split(
             observed_stream_tokens, stream_usage, request_body
@@ -3320,6 +3589,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
             model, route_model, in_tokens + out_tokens, elapsed, route_nodes
         )
         tokens = in_tokens + out_tokens
+        _clear_proxy_progress_log_state(server, request_id)
         _tracker_log(
             server,
             "info",
@@ -3765,6 +4035,68 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
             events = [dict(event) for event in server.console_events]
         self._send_json(200, {"events": events})
 
+    def _handle_proxy_request_cancel(self, request_id: str) -> None:
+        server: _TrackerHTTPServer = self.server  # type: ignore[assignment]
+        if server.accounts is not None and not self._require_role("admin"):
+            return
+        if not _request_proxy_cancel(server, request_id):
+            self._send_json(404, {"error": f"no active proxy for request {request_id!r}"})
+            return
+        self._send_json(200, {"status": "canceled", "request_id": request_id})
+
+    def _finalize_proxy_cancel(
+        self,
+        *,
+        proxy_ctx: _ActiveProxyContext,
+        server: "_TrackerHTTPServer",
+        request_id: str,
+        started: float,
+        model: str,
+        route_model: str,
+        route_nodes: list,
+        api_key: str | None,
+        node_work: list,
+        body: dict,
+        finish_proxy_inflight: Any,
+        observed_stream_tokens: int = 0,
+        stream_usage: dict | None = None,
+    ) -> bool:
+        if not proxy_ctx.cancel_event.is_set():
+            return False
+        elapsed = time.monotonic() - started
+        _clear_proxy_progress_log_state(server, request_id)
+        tokens = observed_stream_tokens
+        if observed_stream_tokens > 0:
+            in_tokens, out_tokens = _stream_billable_split(
+                observed_stream_tokens, stream_usage, body,
+            )
+            tokens = in_tokens + out_tokens
+            self._record_observed_throughput(
+                model, route_model, tokens, elapsed, route_nodes,
+            )
+        _tracker_log(
+            server,
+            "info",
+            "proxy canceled",
+            request_id=request_id,
+            model=model,
+            route_model=route_model,
+            tokens=tokens,
+            elapsed_seconds=round(elapsed, 4),
+            tokens_per_sec=round(tokens / elapsed, 4) if elapsed > 0 and tokens > 0 else 0.0,
+            route=_node_route_summary(route_nodes),
+        )
+        if observed_stream_tokens > 0:
+            in_tokens, out_tokens = _stream_billable_split(
+                observed_stream_tokens, stream_usage, body,
+            )
+            self._bill_completed(
+                api_key, model, in_tokens + out_tokens, node_work,
+                input_tokens=in_tokens, output_tokens=out_tokens,
+            )
+        finish_proxy_inflight()
+        return True
+
     def _handle_registry_wallets(self):
         server: _TrackerHTTPServer = self.server  # type: ignore[assignment]
         if not self._require_role("admin"):
@@ -4532,6 +4864,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
             "model": resolved_name,
             "model_layers_end": required_end,
             "peers": peers,
+            "bytes_per_layer": _preset_bytes_per_layer(preset),
             "model_sources": self._model_sources(
                 resolved_name,
                 preset,

tests/test_node_startup.py

@@ -1118,7 +1118,7 @@ def test_real_model_startup_summary_shows_total_layers(tmp_path, monkeypatch, ca
     assert captured_registration["vram_bytes"] == 6144 * 1024 * 1024
     assert captured_registration["max_loaded_shards"] == 2
     output = capsys.readouterr().out
-    assert "Shard:        layers 0–23; 24 of 24" in output
+    assert "Shard:        layers 0–23 (24 of 24)" in output
     assert "Node ID:      node-test-123" in output
 
 
@@ -1646,6 +1646,106 @@ def test_preset_model_startup_honors_pinned_shard_range(tmp_path, monkeypatch):
         tracker.stop()
 
 
+def test_preset_startup_rejects_pinned_shard_above_memory_budget(tmp_path, monkeypatch):
+    """Pinned layer ranges that exceed the node memory budget fail before model load."""
+    import meshnet_node.startup as startup_mod
+
+    monkeypatch.setattr(
+        startup_mod,
+        "detect_hardware",
+        lambda: {"device": "cpu", "gpu_name": None, "vram_mb": 0, "ram_mb": 8 * 1024},
+    )
+
+    tracker = TrackerServer(model_presets={
+        "big-model": {
+            "layers_start": 0,
+            "layers_end": 39,
+            "hf_repo": "org/big-model",
+            "bytes_per_layer": {"bfloat16": 2 * 1024 * 1024 * 1024},
+        },
+    })
+    tracker_port = tracker.start()
+    tracker_url = f"http://127.0.0.1:{tracker_port}"
+    try:
+        with pytest.raises(ValueError, match="Pinned shard layers 0–39"):
+            run_startup(
+                tracker_url=tracker_url,
+                model="big-model",
+                shard_start=0,
+                shard_end=39,
+                wallet_path=tmp_path / "wallet.json",
+                cache_dir=tmp_path / "shards",
+            )
+    finally:
+        tracker.stop()
+
+
+def test_preset_model_with_hf_repo_loads_torch_backend(tmp_path, monkeypatch, capsys):
+    """Named presets that advertise hf_repo must load TorchNodeServer, not the stub server."""
+    import meshnet_node.startup as startup_mod
+
+    class FakeBackend:
+        total_layers = 16
+
+    torch_calls: list[dict] = []
+
+    class FakeTorchNodeServer:
+        def __init__(self, **kwargs):
+            torch_calls.append(kwargs)
+            self.backend = FakeBackend()
+            self.port = None
+            self.chat_completion_count = 0
+            self.tracker_node_id = None
+
+        def start(self):
+            self.port = 7002
+            return self.port
+
+        def stop(self):
+            pass
+
+    monkeypatch.setattr(
+        startup_mod,
+        "detect_hardware",
+        lambda: {"device": "cpu", "gpu_name": None, "vram_mb": 0, "ram_mb": 16 * 1024},
+    )
+    monkeypatch.setattr(startup_mod, "TorchNodeServer", FakeTorchNodeServer)
+    monkeypatch.setattr(startup_mod, "StubNodeServer", lambda **_kw: (_ for _ in ()).throw(AssertionError("preset with hf_repo must not use StubNodeServer")))
+
+    model_dir = tmp_path / "node-shards" / "tiny-llama"
+    model_dir.mkdir(parents=True)
+    (model_dir / "config.json").write_text('{"num_hidden_layers": 16}')
+    monkeypatch.setattr(startup_mod, "download_shard", lambda *_a, **_kw: model_dir)
+
+    tracker = TrackerServer(model_presets={
+        "tiny-llama": {"layers_start": 0, "layers_end": 15, "hf_repo": "org/tiny-llama-shards"}
+    })
+    tracker_port = tracker.start()
+    tracker_url = f"http://127.0.0.1:{tracker_port}"
+    try:
+        node = run_startup(
+            tracker_url=tracker_url,
+            model="tiny-llama",
+            wallet_path=tmp_path / "wallet.json",
+            cache_dir=tmp_path / "node-shards",
+        )
+        try:
+            assert len(torch_calls) == 1
+            assert torch_calls[0]["model_id"] == "org/tiny-llama-shards"
+            assert torch_calls[0]["cache_dir"] == model_dir
+            output = capsys.readouterr().out
+            assert "Loading real PyTorch model shard..." in output
+            assert "Model ID:     org/tiny-llama-shards" in output
+            network_map = _get_json(f"{tracker_url}/v1/network/map")
+            registered = network_map["nodes"][0]
+            assert registered["hf_repo"] == "org/tiny-llama-shards"
+            assert registered["num_layers"] == 16
+        finally:
+            node.stop()
+    finally:
+        tracker.stop()
+
+
 def test_torch_startup_retries_registration_when_tracker_unreachable(
     tmp_path,
     monkeypatch,

tests/test_real_model_backend.py

@@ -17,6 +17,7 @@ from meshnet_node.model_backend import (
     TensorPayload,
     TorchModelShard,
     _call_layer,
+    _checkpoint_tensor_name_for_model,
     _load_partial_model_from_snapshot,
     _should_partial_materialize_shard,
     _decoder_attention_mask,
@@ -225,7 +226,7 @@ def test_tail_forward_returns_text_completion_from_binary_activations():
         node.stop()
 
 
-def test_full_model_chat_completion_uses_generation_not_single_token_decode():
+def test_full_model_chat_completion_uses_generation_not_single_token_decode(capsys):
     node = TorchNodeServer(backend=_FakeFullBackend())
     port = node.start()
     try:
@@ -237,7 +238,10 @@ def test_full_model_chat_completion_uses_generation_not_single_token_decode():
         req = urllib.request.Request(
             f"http://127.0.0.1:{port}/v1/chat/completions",
             data=payload,
-            headers={"Content-Type": "application/json"},
+            headers={
+                "Content-Type": "application/json",
+                "X-Meshnet-Request-Id": "req-test-123",
+            },
             method="POST",
         )
         with urllib.request.urlopen(req, timeout=5) as resp:
@@ -248,6 +252,10 @@ def test_full_model_chat_completion_uses_generation_not_single_token_decode():
     finally:
         node.stop()
 
+    out = capsys.readouterr().out
+    assert "  [node] processing chat model='fake-model' stream=False max_tokens=7 request_id=req-test-123" in out
+    assert "  [node] chat complete tokens=1 elapsed_s=" in out
+
 
 def test_pipeline_hop_logs_are_suppressed_without_debug(capsys):
     tail_backend = _FakePipelineTailBackend()
@@ -422,7 +430,7 @@ def test_partial_materialize_guard_requires_local_non_full_non_quantized_snapsho
         39,
         total_layers_hint=40,
         uses_quantized_weights=False,
-    ) is False
+    ) is True
     assert _should_partial_materialize_shard(
         str(snapshot_dir),
         4,
@@ -439,6 +447,118 @@ def test_partial_materialize_guard_requires_local_non_full_non_quantized_snapsho
     ) is False
 
 
+def test_checkpoint_tensor_name_remapped_for_text_only_causal_lm():
+    class TextOnlyModel:
+        def __init__(self):
+            self.model = types.SimpleNamespace(layers=[])
+
+    model = TextOnlyModel()
+    assert _checkpoint_tensor_name_for_model(
+        model,
+        "model.language_model.layers.0.mlp.gate.weight",
+    ) == "model.layers.0.mlp.gate.weight"
+    assert _checkpoint_tensor_name_for_model(
+        model,
+        "model.language_model.embed_tokens.weight",
+    ) == "model.embed_tokens.weight"
+
+
+def test_checkpoint_tensor_name_kept_for_multimodal_backbone():
+    class MultimodalModel:
+        def __init__(self):
+            self.model = types.SimpleNamespace(language_model=types.SimpleNamespace())
+
+    model = MultimodalModel()
+    name = "model.language_model.layers.0.mlp.gate.weight"
+    assert _checkpoint_tensor_name_for_model(model, name) == name
+
+
+def test_partial_snapshot_loader_remaps_language_model_checkpoint_keys(tmp_path):
+    snapshot_dir = tmp_path / "snapshot"
+    snapshot_dir.mkdir()
+    (snapshot_dir / "config.json").write_text(json.dumps({
+        "text_config": {"num_hidden_layers": 3},
+    }))
+    (snapshot_dir / "model.safetensors.index.json").write_text(json.dumps({
+        "weight_map": {
+            "model.language_model.layers.1.self_attn.q_proj.weight": "shard-2.safetensors",
+        }
+    }))
+    (snapshot_dir / "shard-2.safetensors").write_bytes(b"stub")
+
+    class FakeModule:
+        def __init__(self):
+            self.to_calls = []
+
+        def to(self, device):
+            self.to_calls.append(device)
+            return self
+
+    class FakeModel:
+        def __init__(self):
+            self.model = types.SimpleNamespace(
+                layers=[FakeModule(), FakeModule(), FakeModule()],
+                rotary_emb=FakeModule(),
+            )
+
+        def tie_weights(self):
+            pass
+
+    class AutoConfigStub:
+        @staticmethod
+        def from_pretrained(model_id):
+            return types.SimpleNamespace(
+                text_config=types.SimpleNamespace(num_hidden_layers=3),
+                get_text_config=lambda: types.SimpleNamespace(num_hidden_layers=3),
+            )
+
+    class AutoModelStub:
+        @staticmethod
+        def from_config(cfg, torch_dtype=None):
+            return FakeModel()
+
+    set_calls = []
+
+    def fake_set_tensor(module, tensor_name, device, value=None, dtype=None):
+        set_calls.append(tensor_name)
+
+    class FakeSafeOpen:
+        def __init__(self, filename, framework, device):
+            self.filename = Path(filename).name
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, exc_type, exc, tb):
+            return False
+
+        def get_tensor(self, tensor_name):
+            return tensor_name
+
+    class UnusedContext:
+        def __enter__(self):
+            return None
+
+        def __exit__(self, exc_type, exc, tb):
+            return False
+
+    _load_partial_model_from_snapshot(
+        AutoConfigStub,
+        AutoModelStub,
+        types.SimpleNamespace(),
+        str(snapshot_dir),
+        1,
+        1,
+        "bf16",
+        "cpu:0",
+        init_empty_weights_fn=UnusedContext,
+        set_tensor_fn=fake_set_tensor,
+        safe_open_fn=FakeSafeOpen,
+    )
+
+    assert set_calls == ["model.layers.1.self_attn.q_proj.weight"]
+
+
 def test_partial_snapshot_loader_materializes_only_assigned_tensors(tmp_path):
     snapshot_dir = tmp_path / "snapshot"
     snapshot_dir.mkdir()

tests/test_tracker_routing.py

@@ -5,6 +5,7 @@ import json
 import threading
 import time
 import urllib.error
+import urllib.parse
 import urllib.request
 
 import pytest
@@ -501,6 +502,100 @@ def test_tracker_logs_stream_progress_before_request_completes():
         node_thread.join(timeout=1.0)
 
 
+def test_tracker_dashboard_can_cancel_inflight_proxy():
+    chunk_sent = threading.Event()
+    release = threading.Event()
+
+    class StreamingChatHandler(http.server.BaseHTTPRequestHandler):
+        def log_message(self, fmt, *args):
+            pass
+
+        def do_POST(self):
+            if self.path != "/v1/chat/completions":
+                self.send_response(404)
+                self.end_headers()
+                return
+            self.rfile.read(int(self.headers.get("Content-Length", 0)))
+            self.send_response(200)
+            self.send_header("Content-Type", "text/event-stream; charset=utf-8")
+            self.end_headers()
+            payload = json.dumps({
+                "choices": [{"delta": {"content": "hello world"}}],
+            }).encode()
+            self.wfile.write(b"data: " + payload + b"\n\n")
+            self.wfile.flush()
+            chunk_sent.set()
+            release.wait(timeout=3.0)
+            self.wfile.write(b"data: [DONE]\n\n")
+            self.wfile.flush()
+
+    node = http.server.HTTPServer(("127.0.0.1", 0), StreamingChatHandler)
+    node_thread = threading.Thread(target=node.serve_forever, daemon=True)
+    node_thread.start()
+    tracker = TrackerServer(heartbeat_timeout=60.0)
+    tracker_port = tracker.start()
+    response = None
+    request_id = None
+    try:
+        _post_json(
+            f"http://127.0.0.1:{tracker_port}/v1/nodes/register",
+            {"endpoint": f"http://127.0.0.1:{node.server_address[1]}",
+             "model": "cancel-proxy-model", "num_layers": 1,
+             "shard_start": 0, "shard_end": 0,
+             "hardware_profile": {}, "score": 1.0},
+        )
+        req = urllib.request.Request(
+            f"http://127.0.0.1:{tracker_port}/v1/chat/completions",
+            data=json.dumps({
+                "model": "cancel-proxy-model",
+                "stream": True,
+                "messages": [{"role": "user", "content": "hi"}],
+            }).encode(),
+            headers={"Content-Type": "application/json"},
+            method="POST",
+        )
+        response = urllib.request.urlopen(req, timeout=3.0)
+        first_line = response.readline()
+        assert first_line.startswith(b"data:")
+        assert chunk_sent.wait(timeout=1.0)
+
+        console = _get_json(f"http://127.0.0.1:{tracker_port}/v1/console")
+        selected = [
+            event for event in console["events"]
+            if event["message"] == "proxy route selected"
+        ]
+        assert selected
+        request_id = selected[-1]["fields"]["request_id"]
+
+        cancel = _post_json(
+            f"http://127.0.0.1:{tracker_port}/v1/proxy/requests/{urllib.parse.quote(request_id, safe='')}/cancel",
+            {},
+        )
+        assert cancel["status"] == "canceled"
+
+        deadline = time.time() + 5.0
+        canceled_events = []
+        while time.time() < deadline:
+            console = _get_json(f"http://127.0.0.1:{tracker_port}/v1/console")
+            canceled_events = [
+                event for event in console["events"]
+                if event["message"] == "proxy canceled"
+                and event["fields"].get("request_id") == request_id
+            ]
+            if canceled_events:
+                break
+            time.sleep(0.05)
+        assert canceled_events
+    finally:
+        release.set()
+        if response is not None:
+            response.close()
+        tracker.stop()
+        node.shutdown()
+        node.server_close()
+        node_thread.join(timeout=1.0)
+
+
 def test_tracker_routes_hf_model_alias_from_quickstart():
     """The documented qwen2.5-0.5b alias resolves a full HF repo registration."""
     tracker = TrackerServer()